基于自适应GA-SVR的中密度纤维板施胶比例辨识方法

Identification Method of Glue Dosing Proportion for MDF Based on Adaptive GA-SVR Algorithm

针对中密度纤维板(MDF)生产过程中施胶比例直接影响产品质量和生产成本的问题,提出了MDF施胶比例的自适应GA-SVR辨识算法。算法采用浮点数与二进制混合编码方式实现优选参数的解空间映射,设计并利用适应度函数完成交叉概率与变异概率的自动调整,通过数代进化实现模型输入参数的优选与支持向量回归参数的优化。为了验证算法的准确性,将辨识算法预测下的施胶比例与比例模型计算出的施胶比例分别用于制板试验,结果表明:基于自适应GA-SVR算法的预测结果与期望值相一致,算法实现了施胶比例的准确预测。

An identification method for glue dosing proportion of MDF was proposed based on adaptive GA-SVR algorithm aiming to solve the problems that glue dosing proportion of MDF directly affects the quality of products and the production cost in the process of MDF production. The space mapping of the optimized parameters achieved by mixing floating-point number with binary encoding, and then the automatic variation of cross-probability and mutation probability was carried out by designing and using fitness function. After evolutions, the model input parameters and support vector regression parameters were optimized. The proposed algorithm was compared with the ratio model in order to verify the accuracy of algorithms. Result shows that the prediction result of adaptive GA-SVR algorithm is consistent with the expectations, and the algorithm could realize the accurate prediction of glue dosing proportion.